JPH04323964A - Picture processing method and device - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image compression/expansion technique for efficiently compressing multivalued images.

0002

2. Description of the Related Art As a compression technique for multivalued images, an ADCT (adaptive discrete cosine transform) compression method has been proposed mainly targeting images such as photographs.

In this compression method, the three primary color signals are converted into Y and Cr.
, Cb, the resolution of the color components Cr and Cb is lowered by subsampling before image compression, and the Y signal, which is the luminance (brightness) component, is compressed with the same resolution.

0004

[Problems to be Solved by the Invention] Conventionally, however, while continuous tone images such as photographic images can be compressed efficiently and with little deterioration, text images cannot be compressed.
The disadvantage was that the edge portions could not be compressed faithfully, resulting in deterioration. Also, the color components of Y, Cb, and Cr are 4:
Subsampling and compressing data to a 1:1 or 4:2:2 ratio results in greater degradation for color characters than for black and white characters. Furthermore, the degree of deterioration of color characters varies depending on the type of color, and all characters cannot be compressed with the same efficiency, which is very inconvenient. In particular, to make characters of all colors legible,
Since the edges of colored characters, which are the most degraded, are preserved, the compression efficiency is extremely reduced.

The present invention has been made in view of the drawbacks of the prior art, and it is an object of the present invention to provide an image processing method and apparatus capable of compressing images while suppressing image deterioration.

[0006]

[Means and Effects for Solving the Problems] In order to solve the above problems, the image processing method of the present invention provides a high degree of preservation according to the importance of each color component when performing different image compression for each color component. It is characterized by performing compression processing.

Furthermore, the image processing apparatus of the present invention is characterized by having means for inputting a plurality of color component data, and a compression means for performing compression processing with different degrees of preservation depending on the importance of the color component data. do.

[0008]

[Embodiment] An embodiment of the present invention performs ADCT compression that compresses color characters with less deterioration without reducing compression efficiency.
By not performing subsampling on the signal component with the largest edge component among the Y, Cr, and Cb signals, and subsampling the rest, the edges of the necessary components of the color character image can be compressed with high resolution. This minimizes the deterioration of all color characters.

Examples of the present invention will be explained in detail below.

FIG. 1 shows an image compression processing circuit.

First, input image data is converted into Y, Cb, and Cr images in a color conversion section. The converted Y, Cb, and Cr data are input to edge detection units 2, 3, and 4, which detect which component among the Y, Cb, and Cr components is changing rapidly in terms of frequency. The detection results of the edge detection units 2, 3, and 4 are sent to the comparison unit 5, which determines whether the frequency is changing rapidly, controls each part, and the code generation unit determines whether the frequency is changing rapidly. Generate the changing image component name, Y, Cb, or Cr as a code. In this embodiment, this judgment can be made for each 8x8 pixel block of ADCT or for each page.Moreover, when the judgment is made for each block, the data is required for decoding, so the compression unit /The decompressor 26 can compress the code generator.

The concrete configuration of the edge detection sections 2, 3, and 4 is as shown in FIG. Thus, a delay of one pixel is performed. This input signal, the output of the line buffer 31, the latch 32,
The outputs of 33 are spatially adjacent as shown in a, b, c, and d in FIG.
The absolute values of six combinations of -c, b-c, b-d, and c-d are calculated, and the maximum values are latched and output for each 8x8 unit to the maximum value determination section 35 and the selection section 36.
is input. The maximum value determination unit 35 determines the maximum value among the maximum values of each of the six combinations, and selects the maximum value from the selection unit 36.
The system is designed to output that value.

By the way, the Y and C output from the color conversion section 1
b, Cr signals are simultaneously processed by sub-sampling units 6, 7, 8.
is input. Here, the component corresponding to 1 of the sampling ratio 4:1:1 is given to the DCT sections 9, 10, and 11, and the remaining 3 parts after subtracting the internal ratio of 1 from the ratio 4 are the components corresponding to the DCT sections 12, 13, and 11.
4 is entered.

FIG. 3 shows 2x2 when performing subsampling.
Indicates pixel components. Therefore, regarding the color components to be subsampled, Ya is input to the DCT section 9, and the DCT section 1
Cba is input to 0, Cra is input to the DCT unit 11, and 8 is subsampled or not subsampled sequentially.
x8 pixels are input. Also, when the edge of the Y component is large, three 8x8
When the Y component of the block is input and the edge of the Cb component is large, the three 8's adjacent to the DCT sections 12, 13, and
When the Cb component of the x8 block is input and the Cr component is large, the Cr components of three adjacent 8x8 blocks are D.
It is controlled by the comparison section 5 so that it is input to the CT sections 12, 13, and 14. DCT sections 9, 10, 11, 12, 13
, 14 are quantized by quantization units 15, 16, 17, 18, 19, and 20, respectively.

The outputs of the quantization units 15, 16, and 17 are input to H encoding units 21, 22, and 23, respectively, which perform Huffman encoding. Furthermore, when the edges of the Y component are large, the outputs of the quantizers 18, 19, and 20 are input to the H encoder 21, and when the edges of the Cb component are large, the outputs of the quantizers 18, 19, and 20 are input to the H encoder 21.
The output of 0 is input to the H encoder 22, and when the Cr component is large, the outputs of the quantizers 18, 19, and 20 are input to the H encoder 23. The inputs of the H encoders 21, 22, and 23 are input to a para/serial converter 24. The para/serial converter 24 rearranges the encoded data depending on whether or not to interleave each color component. In the case of interleaving, the data is rearranged in block order. Y:C
When r:Cb=4:1:1, the order of YYYYCrCb is repeated, and when Y:Cr:Cb=1:4:1, Y
The order is CrCrCrCrCrCb, Y:Cr:Cb=1
:1:4, data configuration is performed by repeating the order of YCrCbCbCbCb. The data rearranged serially in this way is sent out as compressed data.

By the way, the case where image data compressed as described above is decoded will be explained. Compression has been explained in Figure 1, but decompression is possible by completely reversing the direction in which data flows. However, the H encoding units 21, 22, and 23 perform Huffman decoding, and the quantization units 15 to 20 perform inverse quantization. Also, DCT
In sections 9 to 14, coefficients for inverse transformation are set and operated. Further, in the subsampling units 6 to 8, the image is enlarged by the simple iterative method only for the subsampled element components.

Furthermore, during expansion, the edge detection units 2, 3,
4 and comparison unit 5 do not operate, and instead, the code generation unit
Control of parts 50 and 51 in the figure and para/serial conversion unit 24
Block order control is performed. Therefore, during decompression, the compression section/expansion section 26 receives compressed data of the control code, decompresses it, and then delivers it to the code generation section 25.

In this embodiment, the control in the comparing section 5 may be constant over the entire image. In addition, the edge detection units 2, 3, and 4 detect the frequency components of the image and determine whether the edges are large or small.
It can be easily inferred that the sections 12, 13, 14, etc. may be used to detect edges of the required frequency.

[0019] Furthermore, the color components of the image to be compressed are Y, Cr, and C.
Even if R, G, and B are the same instead of b, Y, M, and C
, B, K, Lab, X, Y, Z, etc., of course.

[0020] Also, in this example, the sampling ratio was set to 4:1.
:1, but it goes without saying that 4:2:2 is also applicable.

As described above, according to the above-described embodiment of the present invention, in order to save image information without performing subsampling on the image component whose edge changes the most among each color component, the color To improve the image quality preservation status of only necessary colors at all times without the degree of image quality deterioration varying depending on the component.

[0022]

As described above, the present invention can suppress image deterioration during image compression.

[Brief explanation of drawings]

FIG. 1 is a diagram showing an embodiment of the present invention;

[Figure 2] Edge detection section,

[Figure 3] Diagram explaining subsampling,

[Explanation of symbols]

1 Color conversion section 2, 3, 4 Edge detection part 5 Comparison section 6, 7, 8 Subsampling section 9-14 DCT section 15-20 Quantization section 21-23 H encoding section 24 Para/Series conversion section 25 Code generation section 26 Compression/expansion section 31 Line buffer 32, 33 latch 34 Difference calculation section 35 Maximum value judgment section 36 Selection section

Claims (5)

[Claims] 1. An image processing method that performs different image compression processing for each color component, the image processing method being characterized in that compression processing with a high degree of preservation is performed according to the importance of each color component. 2. The image processing method according to claim 1, wherein the degree of importance is determined for each pixel. 3. The image compression method is ADCT (Ad
aptive Discrete Cosine
(Transform) An image processing method, wherein the compression processing with a high degree of preservation is processing in which subsampling is not performed in subsampling. 4. An image processing apparatus comprising: input means for inputting a plurality of color component data; and compression means for performing compression processing with different degrees of preservation depending on the importance of the color component data. 5. The image processing apparatus according to claim 4, further comprising determining means for determining the degree of importance for each pixel.